MaGIC Signature Scoring Tool
MaGIC Signature Scoring Tool
Welcome to the Signature Scoring Tool by the Molecular and Genomics Informatics Core (MaGIC).
Score every sample against any number of gene-set signatures, producing a signature × sample matrix that flows directly into the MaGIC Survival and Correlation tools.
How to Use This Tool
- Open the Data & Scoring tab. Use the built-in demo data, or switch on 'Upload custom data' to load a normalized expression matrix (genes × samples, log-scale preferred) and sample metadata.
- Choose a signature source. Built-in demo signatures, a GMT file, MSigDB collections, or pasted custom signatures.
- Pick a scoring method. GSVA, ssGSEA, AUCell, or singscore, with method-specific parameters.
- Click Run Scoring. The Loaded Signatures and Score Matrix sub-tabs populate (each table has its own download button), and the Score Heatmap and Distributions tabs appear.
- Explore & download. View the annotated signature × sample heatmap and per-signature distributions; download the score matrix in the orientation Survival/Correlation expect, plus your plots.
Which scoring method should I use?
GSVA
Non-parametric, sample-rank based enrichment. Scores are computed relative to the rest of the cohort, making it ideal for stable comparative analysis across samples. Use the Gaussian kcdf for continuous data (log-CPM/TPM/VST) and Poisson for integer counts.
AUCell
Rank-based area-under-the-curve scoring. Robust to dropouts and varying detection depth because it only considers whether signature genes fall in the top fraction of each sample's ranking.
ssGSEA
Single-sample GSEA. Produces a direct per-sample enrichment score for each signature, computed independently per sample. Optional normalization rescales scores across the dataset.
singscore
Rank-based directional single-sample scoring. Choose whether each signature is interpreted as up-regulated, down-regulated, or undirected (bidirectional).
Required Input Data Formats
Expression Matrix
- File format: CSV or TSV
- Rows: Genes (one gene per row)
- Columns: Samples (one sample per column)
- First column: Gene identifiers (symbols, matched to gene sets)
- Values: normalized, log-scale preferred (e.g. log2-CPM, VST) — the same schema as the MaGIC QC tool's output
Gene, Sample1, Sample2 BRCA1, 6.5, 7.1 TP53, 8.2, 7.9
Sample Metadata
- File format: CSV or TSV
- Rows: Samples (one sample per row)
- First column: Sample names — must match matrix column names exactly
- Additional columns: Categorical or numeric metadata (e.g. Group, Sex, Batch)
Sample, Group, Sex Sample1, Control, Male Sample2, Treated, Female
1. Data
Pre-loaded demo: 3 expression modules + background genes across 30 samples in 3 treatment groups, with matching built-in signatures.
2. Gene set signatures
3. Scoring method
Genes are matched against the genes in your expression matrix; signatures with fewer than 2 matched genes are skipped at scoring time.
Signatures (rows) × samples (columns) — the matrix consumed by downstream tools.